Demystifying data centres

With three-quarters of all Australian companies hosting their servers at their offices it's time to shake up the status quo.

Few would dispute the notion that the world has witnessed an information technology revolution in recent years, which has resulted in greater demands on technology and an ever-increasing amount of data being produced, stored and transferred every day. Despite this surge in data, companies all over Australia continue to accommodate their own data centres on-site, running the risk of over-expenditure and burdensome inefficiencies.

A recent industry report suggests that as many
 as three-quarters of all Australian companies host their servers at their offices, whether tucked away in a corner or by leasing additional floor space in the building. While this might seem like an easy, economical solution that keeps data close at hand and allows companies to expand their IT functions as they grow, the reality is that the decision to do so carries a much greater cost than is understood.

Total cost of occupancy considerations

Perhaps the most obvious cost factor associated with hosting the company data centre on-site is the lease itself. Some companies choose “premium” buildings to ensure they receive the minimum power and cooling they need for their IT equipment to run smoothly. However, the rent in premium buildings is typically high, which makes it difficult to calculate a true total cost of occupancy (TCO) that factors in both employee office space and data centre space.

There are also several critical technical issues that need to be considered with an on-site data centre. For example, data centres require approximately 1 kilowatt (kW) of air conditioning for every 1 kW of heat inside the rack in order to ensure the technology runs efficiently. As a result, these companies face significant additional costs in terms of power and cooling, often accounting for up to half the cost of running the entire office.

Further, most office buildings simply aren’t equipped to accommodate data centres. For example, they have centralised cooling systems installed to maintain a consistent temperature throughout the building for human occupation, and they have a centralised low-density power supply that is not designed to support large volumes of technology in a limited space. In the rare cases where commercial buildings can support stand-alone cooling systems for data centres, they typically lack the necessary generators to provide resilience that assures system availability in the event of power outages
or downtime.

Structurally, office buildings aren’t designed to house data centres either—they lack the necessary slab-to-slab heights, for example. Companies tend to stack critical equipment on low-height shelving units, and the same are cooled via low-height raised floors, which can easily result in overheating. Commercial property also doesn’t offer adequate structural loading to accommodate the ever-increasing physical weight of a typical server room or data centre, which can reach a critical mass of up to 1,400 kilograms in some cases.

Recognising security risks

Organisations that house their data centres in commercial buildings also face a host of security risks because their offices may not be safeguarded against theft, damage and sabotage, for example. Given that commercial landlords are under no obligation to meet service level agreements (SLAs) in this area, there is very little companies can do to protect themselves against any of these concerns.

There is no doubt that many companies will continue to host their data centres in-house; however, by maintaining the status quo, these organisations not only face the ongoing challenges outlined above, they will also struggle to meet uptime and reliability requirements, which are an absolute given in today’s data-charged world.

The combined hard and soft costs of an in-house data centre can be astronomical, so it is important to properly calculate the TCO before deciding whether to continue with this
model or outsource to a secure, cost-effective and reliable alternative. To do this, a rigorous occupancy cost classification system such as the International Total Occupancy Cost Code should be used to benchmark the available options. By capturing the full impact of capital and operating expense items over the likely term of the occupation, a more complete cost comparison emerges.

Understanding the options

There are a number of models that can provide viable solutions to companies and it is essential to understand all the costs surrounding each before deciding on the most appropriate facility.

Build-your-own

This option can be effective for companies with larger requirements, but there is a tendency to over-provide for smaller clients, which leads to stranded capacity. Moreover, bespoke designs tend to be more expensive and less power efficient than commercially operated data centres.

Colocation

A viable option if not much space is required, and the low entry costs and scalability are very attractive to organisations that require less than 20 racks. The option to bundle power, networking and services into one contract is also convenient for smaller users. Some costs such as network cross-connects, and cleaning and power usage above the contracted capacity are typically not included.

Turnkey lease

A solution very low on hidden costs, offering clients a secure, reliable facility where rent is typically charged by the power used rather than the floor space occupied.

There are many quantitative and qualitative benefits to be gained by using dedicated data centre facilities over in-house data centres. The implicit assumption made by companies that they cannot afford to use a dedicated facility must be challenged by comparing total occupancy costs over the expected term of their requirements to ensure they are getting the best value in this increasingly data driven world.

Peter Wolsey is the director of corporate development, Australia for Digital Realty